ECE291 Computer Engineering II Lecture 6 & Lecture 7

1. ECE291Computer Engineering IILecture 6 & Lecture 7 Dr. Zbigniew Kalbarczyk University of Illinois at Urbana- Champaign

2. Outline • Program organization • MASM directives • Multiplication • Division ECE291

3. Program Organization • Create block structure and/or pseudocode on paper to get a clear concept of program control flow and data structures • Break the total program into logical procedures/macros • Use jumps, loops, etc. where appropriate • Use descriptive names for variables • noun_type for types • nouns for variables • verbs for procedures/functions ECE291

4. Debugging Hints • Good program organization helps • Programs do not work the first time • Strategy to find problems • Use DEBUG breakpoints to check program progress • Use COMMENT to temporarily remove sections of code • "print" statements announce milestones in program • Test values/cases • Try forcing registers/variables to test output of a procedure • Use "print" statements to display critical data • Double-check your own logic (Did you miss a special case?) • Try a different algorithm, if all else fails... ECE291

5. MASM Directives • General • TITLE my_program.asm • Includes drive:\path\filename • Definitions • DB define byte (8bits) • DW define word (16 bits) • DD define doubleword (32 bits) • EQU names a constant • Labels ECE291

6. MASM Directives • Macros • Instead of using procedures, which require both stack and time resources, macros are fast and flexible • Advantages: • speed; no call instruction • readability - easier to understand program function • Drawbacks - • space using the MACRO multiple times duplicates the code • tricky to debug (in particular when you have nested MACROs) • Procedures • “name” PROC NEAR/FAR • ENDP ECE291

7. MASM Directives (cont.) • References to procedures • EXTERN “name” NEAR/FAR/BYTE/WORD • PUBLIC “ name” • Segment definition • SEGMENT “name” PUBLIC/STACK • ENDS • Segment register “Hooks” • ASSUME CS:CSEG; DES:CSEG; SS:STACK ECE291

8. Example Program Structure TITLE ECE291:MPXXX COMMENT * In this MP you will develop program which take input from the keyboard………… * ;====== Constants================================================= ;ASCII values for common characters CR EQU 13 LF EQU 10 ESCKEY EQU 27 ;====== Externals================================================= ; -- LIB291 Routines extrn dspmsg:near, dspout:near, kbdin:near extrn rsave:near, rrest:near, binasc:near ECE291

9. Example Program Structure (cont.) ;==== LIBMPXXX Routines (Your code will replace calls to these functions) extrn LibKbdHandler:near extrn LibMouseHandler:near extrn LibDisplayResult:near extrn MPXXXXIT:near ;====== Stack ===================================================== stkseg segment stack ; *** STACK SEGMENT *** db 64 dup ('STACK ') ; 64*8 = 512 Bytes of Stack stkseg ends ;====== Begin Code/Data============================================ cseg segment public 'CODE' ; *** CODE SEGMENT *** assume cs:cseg, ds:cseg, ss:stkseg, es:nothing ECE291

10. Example Program Structure (cont.) ;====== Variables================================================= inputValid db 0 ; 0: InputBuffer is not ready ; 1: InputBuffer is ready ;-1: Esc key pressed operandsStr db 'Operands: ','$' OutputBuffer db 16 dup(?),'$' ; Contains formatted output ; (Should be terminated with '$') MAXBUFLENGTH EQU 24 InputBuffer db MAXBUFLENGTH dup(?),'$' ; Contains one line of ; user input include graphData.dat ; data PUBLIC OutputBuffer, inputValid, operandsStr PUBLIC graphData ECE291

11. Example Program Structure (cont.) ;====== Procedures =========================================== KbdHandler PROC NEAR <Your code here> KbdHandler ENDP MouseHandler PROC NEAR <Your code here> MouseHandler ENDP DisplayResult PROC NEAR <Your code here> DisplayResult ENDP ECE291

12. Example Program Structure (cont.) ;====== Main Procedure ======================================== MAIN PROC FAR MOV AX, CSEG ;Use common code and data segment MOV DS, AX MOV AX, 0B800h ;Use extra segment to access video screen MOV ES, AX <here comes your main procedure> CALL MPXXXXIT ; Exit to DOS MAIN ENDP CSEG ENDS END MAIN ECE291

13. Multiplication • The product after a multiplication is always a double-width product, e.g, • if we multiply two 16-bit numbers , they generate a 32-bit product • unsigned: (216 - 1) * (216 - 1) = (232 - 2 * 216 + 1 < (232 - 1) • signed: (-215) * (-215) = 230 < (231 - 1) • overflow cannot occur • Modification of Flags • Most flags are undefined after multiplication • O and C flags clear to 0 if the result fit into half-size register • e.g., if the most significant 16 bits of the product are 0, both flags C and O clear to 0 ECE291

14. Multiplication (cont.) • Two different instructions for multiplication • MULMultiply unsigned • IMUL Integer Multiply (2’s complement) • Multiplication is performed on bytes, words, or double words • Which operation to perform depends on the size of the multiplier • The multiplier can be any register or any memory location mul cx ; AX * CX (unsigned result in DX--AX);imul BYTE PTR [si] ; AX * [word contents of memory location ; addressed by SI] (signed productin DX--AX) ECE291

15. Multiplication(16 bit) The use of the AX (and DX) registers is implied!!!!! Multiplicand AX Multiplier (16-bit register, 16-bit memory variable) DX, AX = PRODUCT (High word in DX : Low word in AX) ECE291

16. Multiplication (cont.) • 8086/8088 microprocessors do not allowto perform immediate multiplication • 80286, 80386, and 80486 allow the immediate multiplication by using a special version of the multiply instruction • Immediate multiplication must be signed multiplication and contains three operands • 16-bit destination register • register or memory location that contains 16-bit multiplicand • 8-bit or 16-bit immediate data used as a multiplier mul cx, dx, 12h ;multiplies 12h * DX and leaves ;16-bit signed product in CX ECE291

17. Multiplication • 8-bit multiplication Multiplicand AL Multiplier (8-bit register, 8-bit memory variable) AX PRODUCT • 32-bit multiplication Multiplicand EAX Multiplier (32-bit register, 32-bit memory variable) EDX, EAX PRODUCT (High word in EDX : Low word in EAX) • 32-bit multiplication is available only on 80386 and above ECE291

18. Binary Multiplication • Long Multiplication is done through shifts and additions • This works if both numbers are positive • To multiply a negative numbers, the CPU will store the sign bits of the numbers, make both numbers positive, compute the result, then negate the result if necessary 0 1 1 0 0 0 1 0 (98) x 0 0 1 0 0 1 0 1 (37) ------------------------- 0 1 1 0 0 0 1 0 0 1 1 0 0 0 1 0 - - 0 1 1 0 0 0 1 0 - - - - - (3626) ECE291

19. Division • X / Y = Q; R X Dividend Y Divisor Q Quotient R Remainder Note: Remainder has the same sign as X (Dividend) Examples (Signed Integers) X / Y Q R 9 / 4 2 1 -9 / 4 -2 -1 9 / -4 -2 1 -9 / -4 2 1 ECE291

20. Division (cont.) • Two different instructions for division • DIVDivision unsigned • IDIV Integer Division (2’s complement) • Division is performed on bytes, words, or double words • Which operation to perform depends on the size of the divisor • The dividend is always a double-width dividend that is divided by the operand (divisor) • The divisor can be any register or any memory location ECE291

21. Division(32-bit/16-bit) The use of the AX (and DX) registers is implied!!!!! Dividend DX, AX (high word in DX, low word in AX) Divisor (16-bit register, 16-bit memory variable) Quotient AX Remainder DX ECE291

22. Division (cont.) • 16-bit/8-bit Dividend AX Divisor (8-bit register, 8-bit memory variable) Quotient AL Remainder AH • 64-bit/32-bit Dividend EDX, EAX (high double word in EDX, low double word in EAX) Divisor (32-bit register, 32-bit memory variable) Quotient EAX Remainder EDX • Available on 80386 and above ECE291

23. Division (cont.) • Division of two equally sized words • Prepare the dividend • Unsigned numbers: move zero into high order-word • Signed numbers: use signed extension (implicitly uses AL, AX, DX registers) to fill high-word with once or zeros • CBW(convert byte to word) AX = xxxx xxxx snnn nnnn (before) AX = ssss ssss snnn nnnn (after) • CWD(convert word to double) DX:AX = xxxx xxxx xxxx xxxx snnn nnnn nnnn nnnn (before) DX:AX = ssss ssss ssss ssss snnn nnnn nnnn nnnn (after) • CWDE(convert double to double-word extended) - 80386 and above ECE291

24. Division (cont.) • Flag settings • none of the flag bits change predictably for a division • A division can result in two types of errors • divide by zero • divide overflow (a small number divides into a large number), e.g., 3000 / 2 • AX = 3000; • Devisor is 2 => 8 bit division is performed • Quotient will be written to AL => but 1500 does not fit into AL • consequently we have divide overflow • in both cases microprocessor generates interrupt (interrupts are covered later in this course) ECE291

25. Division (Example) Division of the byte contents of memory NUMB by the contents of NUMB1 Unsigned MOV AL, NUMB ;get NUMB MOV AH, 0 ;zero extend DIV NUMB1 MOV ANSQ, AL ;save quotient MOV ANSR, AH ;save remainder Signed MOV AL, NUMB ;get NUMB CBW ;signed-extend IDIV NUMB1 MOV ANSQ, AL ;save quotient MOV ANSR, AH ;save remainder ECE291

26. Division (cont.) • What do we do with remainder after division? • use the remainder to round the result • drop the remainder to truncate the result • if the division is unsigned, rounding requires that remainder is compared with half the divisor to decide whether to round up the quotient • e.g., sequence of instructions that divide AX by BL and round the result DIV BL ADD AH, AH ;double remainder CMP AH, BL ;test for rounding JB NEXT INC AL NEXT: ECE291